Spectroscopic Characterization of Copper(II) Binding to the

Princess Alexandra Hospital, Ipswich Road, Brisbane, QLD 4102, Australia ... Using a range of spectroscopic techniques we show that Cu(II) is coor...
8 downloads 0 Views 182KB Size
Download PDF
Published on Web 07/04/2006

Spectroscopic Characterization of Copper(II) Binding to the Immunosuppressive Drug Mycophenolic Acid Christopher E. Jones,*,†,‡ Paul J. Taylor,| Alastair G. McEwan,†,‡ and Graeme R. Hanson†,§ Contribution from the Centre for Metals in Biology, School of Molecular and Microbial Sciences, and Centre for Magnetic Resonance, The UniVersity of Queensland, St Lucia, Queensland, 4072, and Departments of Medicine, The UniVersity of Queensland, and Clinical Pharmacology, Princess Alexandra Hospital, Ipswich Road, Brisbane, QLD 4102, Australia Received November 10, 2005; Revised Manuscript Received April 24, 2006; E-mail: [email protected]

Abstract: Mycophenolic acid (MPA) is a drug that has found widespread use as an immunosuppressive agent which limits rejection of transplanted organs. Optimal use of this drug is hampered by gastrointestinal side effects which can range in severity. One mechanism by which MPA causes gastropathy may involve a direct interaction between the drug and gastric phospholipids. To combat this interaction we have investigated the potential of MPA to coordinate Cu(II), a metal which has been used to inhibit gastropathy associated with use of the NSAID indomethacin. Using a range of spectroscopic techniques we show that Cu(II) is coordinated to two MPA molecules via carboxylates and, at low pH, water ligands. The copper complex formed is stable in solution as assessed by mass spectrometry and 1H NMR diffusion experiments. Competition studies with glycine and albumin indicate that the copper-MPA complex will release Cu(II) to amino acids and proteins thereby allowing free MPA to be transported to its site of action. Transfer to serum albumin proceeds via a Cu(MPA)(albumin) ternary complex. These results raise the possibility that copper complexes of MPA may be useful in a therapeutic situation.

1. Introduction.

Mycophenolic acid (MPA, Chart 1) is a frontline immunosuppressant used for the prophylaxis of rejection following solid organ transplantation. By the late 1990s, over 75% of renal transplant patients in the US were receiving mycophenolate mofetil (MMF), the prodrug of MPA.1,2 MPA exerts its clinical effect by binding to the active site in the enzyme inosine monophosphate dehydrogenase (IMPDH). This binding inhibits the proliferation of human T and B lymphocytes in two ways. First, IMPDH catalyzes the de novo pathway of guanosine nucleotide synthesis, a pathway that T and B lymphocytes are almost solely dependent upon. Second, MPA inhibits the type II isoform of IMPDH that is expressed in activated lymphocytes, rather than the type I “housekeeping” isoform. Hence MPA is able to selectively inhibit lymphocytes over other cell types.3,4 IMPDH catalyzes the nicotinamide adenine dinucleotide (NAD) dependent oxidation of inosine-5′-monophosphate (IMP) to xanthosine-5′-monophosphate (XMP). MPA binds to the nicotinamide portion of the NAD binding site after NADH release has occurred but before XMP has been produced. The crystal structure of MPA bound IMPDH shows MPA bound along with †

Centre for Metals in Biology. School of Molecular and Microbial Sciences. § Centre for Magnetic Resonance. | Departments of Medicine. ‡

(1) (2) (3) (4)

Behrend, M.; Braun, F. Drugs 2005, 65, 1037-1050. Takemoto, S. K. Transplant. Proc. 2002, 34, 1632-1634. Allison, A. C.; Eugui, E. M. Immunopharmacology 2000, 47, 85-118. Allison, A. C. Lupus 2005, 14 Suppl 1, s2-8.

9378

9

J. AM. CHEM. SOC. 2006, 128, 9378-9386

Chart 1. a

a (Panel A) Schematic representation of MPA showing numbering scheme used. (Panel B) Ball and stick model of MPA.

an IMP intermediate.5 All functional groups of MPA are required for binding to IMPDH, including the C2′-C3′ (see Chart 1) double bond which keeps the hexanoic acid side chain in a bent conformation. In addition to IMPDH inhibition, MPA has additional therapeutic effects by inhibiting inducible nitric oxide synthase, an enzyme linked to renal allograft rejection episodes.4,6 (5) Sintchak, M. D.; Fleming, M. A.; Futer, O.; Raybuck, S. A.; Chambers, S. P.; Caron, P. R.; Murcko, M. A.; Wilson, K. P. Cell 1996, 85, 921-30. (6) Senda, M.; DeLustro, B.; Eugui, E.; Natsumeda, Y. Transplantation 1995, 60, 1143-8. 10.1021/ja057651l CCC: $33.50 © 2006 American Chemical Society

Copper(II) Binding to Mycophenolic Acid

Despite many therapeutic benefits, optimal use of MPA has been hampered by a range of side effects. Gastrointestinal (GI) effects are prevalent and in some cases are severe enough to require either cessation or dose changes. Dose changes are known to increase the chance of organ rejection.1 Oral bioavailability of MPA is increased by administration of the ester prodrug MMF. MMF undergoes rapid hydrolysis to MPA followed by metabolism to MPA glucuronide.7,8 The rapid hydrolysis of MMF occurs high up in the GI tract and is thought to be partly responsible for the observed GI side effects.9 In an attempt to combat the adverse effects, an enteric coated formulation of sodium mycophenolate (EC-MPS) was developed.10 This formulation is designed to dissolve lower down the GI tract at low pH values, hence improving upper GI tolerability. Some cases have indeed shown a remarkable improvement in GI events after the introduction of EC-MPS.11 Clinical studies have shown this formulation to be therapeutically equivalent to MMF. Despite the enteric coat, the studies have also shown that the overall incidence of adverse events and GI side effects is largely similar between patients on MMF and those on EC-MPS; however the GI severity tended to be lower in the EC-MPS group.12-14 Gastrointestinal intolerability is recognized as a significant side effect in the use of many drugs. In particular, the GI effects of aspirin and other nonsteroidal anti-inflammatory drugs (NSAID) are well documented.15,16 The potent NSAID, indomethacin, has dose-dependent GI side effects in humans and has fatal outcomes if used in dogs.15 For many years the use of copper to inhibit inflammation has been reported.15,17,18 Moreover, many reports show that copper complexed NSAIDs are often more effective than the copper salt or the uncomplexed NSAID alone.17 Significantly, many Cu-NSAID complexes also have greater GI tolerability than the parent NSAID.15 The GI sparing ability of copper-NSAID complexes is highlighted by copper-bound indomethacin,19 which has a therapeutic action equivalent to free indomethacin but results in much lower GI damage.20 Consequently, Cu-indomethacin has found use as an anti-inflammatory in veterinary applications.15 Copper is the third most abundant transition metal in biological systems (after iron and zinc), and although it is toxic at high concentrations, it is nevertheless attractive for use in therapeutic applications as organisms have very sophisticated mechanisms to sequester (7) Bullingham, R. E.; Nicholls, A.; Hale, M. Transplant. Proc. 1996, 28, 9259. (8) Johnson, A. G.; Rigby, R. J.; Taylor, P. J.; Jones, C. E.; Allen, J.; Franzen, K.; Falk, M. C.; Nicol, D. Clin. Pharmacol. Ther. 1999, 66, 492-500. (9) Holt, C. D.; Sievers, T. M.; Ghobrial, R. M.; Rossi, S. J.; Goss, J. A.; McDiarmid, S. V. BioDrugs 1998, 19, 373-384. (10) Rihs, G.; Papageorgiou, C.; Pfeffer, S. Acta Crystallogr. 2000, C56, 432433. (11) Suwelack, B.; Gabriels, G.; Volmer, S.; Hillebrand, U.; Hohage, H.; Pohle, T. Transplantation 2005, 79, 987-988. (12) Salvadori, M.; Holzer, H.; de Mattos, A.; Sollinger, H.; Arns, W.; Oppenheimer, F.; Maca, J.; Hall, M.; Groups, T. E. B. S. Am. J. Transplant. 2004, 4, 231-236. (13) Sollinger, H. Transplant. Proc. 2004, 36, 517s-520s. (14) Budde, K.; Glander, P.; Diekmann, F.; Dragun, D.; Waiser, J.; Fritsche, L.; Neumayer, H. H. Transplant. Proc. 2004, 36, 524s-527s. (15) Weder, J. E.; Dillon, C. T.; Hambley, T. W.; Kennedy, B. J.; Lay, P. A.; Biffin, J. R.; Regtop, H. L.; Davies, N. M. Coord. Chem. ReV. 2002, 232, 95-126. (16) Rodriguez, L. Lancet 1994, 343, 769-772. (17) Sorenson, J. R. J. Inflammatory Diseases and Copper, 1st ed.; Humana Press: Clifton, NJ, 1982. (18) Sorenson, J. R. J. In A Physiological Basis for Pharmacological ActiVities of Copper Complexes, A hypothesis; Sorenson, J. R. J., Ed. The Biology of Copper Complexes; Humana Press: Clifton, NJ, 1987. (19) Weder, J. E., et al. Inorg. Chem. 1999, 38, 1736-1744. (20) Dillon, C. T.; Hambley, T. W.; Kennedy, B. J.; Lay, P. A.; Zhou, Q.; Davies, N. M.; Biffin, J. R.; Regtop, H. L. Chem. Res. Toxicol. 2003, 16, 28-37.

ARTICLES

and transport the metal. In particular, specific metallochaperones bind copper ions in sites that inhibit deleterious cellular interactions and also transfer the metal to specific copper-requiring enzymes and proteins.21,22 Although the pathophysiology of GI intolerance of NSAIDs and MPA is very complex and not clearly understood, it is thought that one of the mechanisms underlying GI toxicity is a direct interaction between the drug and the gastric phospholipids. An electrostatic interaction between negatively charged carboxyl groups on the drug and the positively charged quaternary ammonium group of phosphatidylcholine has been suggested as a mechanism behind NSAID gastropathy.23 This interaction appears to reduce the ability of surface-active phospholipids to form a hydrophobic protective layer.23 Such a topical effect may also account for some of the GI effects of MPA, and the effects would only be exacerbated by the rapid hydrolysis of MMF to free-acid MPA. Copper binding to indomethacin occurs via the carboxylate group.19 The complexation most likely inhibits the ability of the carboxylate to bind to lipid headgroups which may partially explain the observed reduction in GI toxicity.23 Indomethacin is a nonselective inhibitor of cyclooxygenase (COX). Two isoforms of COX exist; COX-1 has a number of roles including controlling the gastric mucosa, while COX-2 produces the prostaglandins involved in inflammation and mitogenesis. Indomethacin-induced inhibition of COX-1 is thought to play a role in GI ulcerogenic activity. Currently it is unclear if copper-indomethacin plays a role in the COX enzyme system; however, uptake of intact copper-indomethacin complexes has been reported, and the complex remains a potent anti-inflammatory.15,24 Additionally, the copper-indomethacin complex displayed superoxide dismutase (SOD) activity, which was suggested to contribute to the lower ulcerogenic activity of the copper complex compared to indomethacin alone.20 SOD is available in the gastric mucosa of the GI tract and is thought to be involved in protecting the gastric and duodenal mucosa from damaging species.25 Thus, copper-indomethacin may reduce gastric damage via a number of mechanisms. Based on the observation that enteric-coated MMF results in reduced severity of adverse GI events, combined with the successful use of copper to combat the GI effects of indomethacin and related NSAIDs, we investigated the potential use of this metal ion to coordinate MPA. We have characterized the binding of copper to MPA using a range of spectroscopic techniques including NMR, EPR, and mass spectrometry. Additionally, the affinity of Cu(II) for MPA has been estimated. Further, as MPA is highly bound to serum albumin (97%),1 and the protein has numerous Cu(II) binding sites, we have investigated the interaction of copper-bound MPA with albumin. The copper-MPA complex formed may be beneficial as an immunosuppressant having reduced GI toxicity. 2. Experimental 2.1. Materials. MPA (E-6-(1,3-dihydro-7-hydroxy-5-methoxy-4methyl-1-oxoisobenzofuran-6-yl)-4-methyl-4-hexanoic acid) was obtained from the Sigma Chemical Co. (St. Louis, MO) and used without (21) Harrison, M. D.; Jones, C. E.; Solioz, M.; Dameron, C. T. Trends Biochem. Sci. 2000, 25, 29-32. (22) Tottey, S.; Harvie, D. R.; Robinson, N. J. Acc. Chem. Res. 2005, 38, 775783. (23) Giraud, M.-N.; Motta, C.; Romero, J. J.; Bommelaer, G.; Lichtenberger, L. M. Biochem. Pharmacol. 1999, 57, 247-254. (24) Lee, S. Y.; Belmonte, A. A. J. Pharm. Sci. 1994, 83, 1107-9. (25) Klinowski, E.; Broide, E.; Varsano, R.; Eshchar, J.; Scapa, E. Eur. J. Gastroenterol. Hepatol. 1996, 8, 1151-5. J. AM. CHEM. SOC.

9

VOL. 128, NO. 29, 2006 9379

Jones et al.

ARTICLES further purification. Deuterated water and methanol were obtained from Cambridge Isotopes. During titrations, copper was added as small aliquots from a freshly prepared stock solution of CuCl2‚2H2O. Where pH adjustments were required K2HPO4/KH2PO4 or KOH was used. All chemicals and solvents were AR grade unless otherwise stated and were used without further purification. All experiments were conducted in a range of differing solvent concentrations (e.g., 100% MeOH to 4% MeOH in water) where possible, and the solvent conditions were found to not interfere with the results. 2.2. UV/visible Spectroscopy. UV/visible electronic absorption spectra were obtained on a Cary 3010 spectrometer, using a 1 cm path length cell. All spectra were acquired over a wavelength range from 220 to 1000 nm. Analysis of the λmax of the Cu(II) d-d transitions was undertaken using the method of Billo,37 updated by Prenesti and co-workers.38 Briefly, an estimate of the absorption maximum can be calculated with the equation:

and were based on those of Makara et al.29 Diffusion coefficients for apo- and copper-bound MPA in d4-MeOH were determined using the DOSY program supplied in the Topspin software (Bruker). Pulse field gradient NMR spectra were acquired using a bipolar gradient, stimulated echo sequence. Diffusion coefficients were measured by incrementing the amplitude of the field gradient pulses over 32 steps (0.8-32 G/cm). The duration of the field gradient pulse (10ms) and the diffusion time (200ms) were held constant for all experiments. The spectra were recorded with 32 scans in a 2D mode, and a relaxation delay of 2 s between scans. The gradient strength was calibrated using the diffusion of residual water in a 100% D2O sample (∼1.9 × 10-9 m2 s-1).30 The Bruker T1/T2 relaxation routine was used for all data analysis. Peak intensities were fitted to an exponential decay using the SimFit program within the Topspin software to provide estimates of the diffusion constant (D). The hydrodynamic radius (rs) of MPA and copper-bound MPA were estimated from the Stokes-Einstein equation:

λmax ) 103/ΣiniνI

D ) kT/6πηrs

(1)

where ni is the number of each equatorial donor group (1e n e 4) and νi represents the individual contribution of each equatorial ligand to the ligand field of the complex. For a carboxylate oxygen νi ) 0.353 ( 0.008 µm-1; water oxygen ) 0.296 ( 0.006 µm-1, and hydroxide oxygen ) alcoholate oxygen ) 0.39 ( 0.03 µm-1.37,38 2.3. Electron Paramagnetic Resonance Spectroscopy (EPR). Continuous wave EPR spectra at either ∼9.4 GHz (X-Band) or ∼4 GHz (S-Band) were obtained on a Bruker Elexsys E500 spectrometer operated with Bruker Xepr software. Spectra were obtained using either a super high-Q or optical cavity (X-band) or a flexline resonator (Sband). Calibration of the magnetic field was achieved using an ER035 gaussmeter. The microwave frequency was calibrated with an EIP548B microwave frequency counter. A Eurotherm B-VT-2000 variable temperature controller provided stable temperatures of ∼140 K. Spectra were routinely baseline corrected using polynomial functions and smoothed using Fourier filtering available in the Xepr software. Simulations of the EPR spectra were performed using XSophe (version 1.1.4) running on a Linux workstation. Spectra were routinely simulated using matrix diagonalization for the analysis of randomly oriented EPR samples, starting with experimentally derived g and A matrices.26 2.4. Nuclear Magnetic Resonance Spectroscopy (NMR). All NMR spectra were obtained on a Bruker Avance 500 MHz spectrometer using a 5 mm triple-resonance, z-gradient probe. MPA samples were prepared in either d4-MeOH or 90% D2O/10% d4-MeOH. All spectra were acquired at 303 K. If required, the residual water signal was suppressed using either a low power presaturation pulse or a W5 watergate sequence.27 Proton spectra were generally acquired over a 10 ppm spectral width, with 32K real points, and 13C spectra were acquired over a 200 ppm spectral width. Data were acquired and processed using Bruker Topspin software running on Linux workstations. Generally, the one-dimensional spectra were processed using a π/2 shifted sinesquared window function. Spectra were referenced to the residual methanol signal at 3.30 ppm (for 1H spectra) and at 58 ppm (for 13C spectra). Resonance assignments for apo-MPA were determined using 1D 1H and 13C-DEPT experiments and standard chemical shift tables28 (26) Hanson, G. R.; Gates, K. E.; Noble, C. J.; Griffin, M.; Mitchell, A.; Benson, S. J. Inorg. Biochem. 2004, 98, 903-916. (27) Lui, M.; Mao, X.; Ye, C.; Huang, H.; Nicholson, J. K.; Lindon, J. C. J. Magn. Res. 1998, 132, 125-129. (28) Abraham, R. J.; Fisher, J.; Loftus, P. Introduction to NMR spectroscopy; John Wiley and Sons: New York, 1988. (29) Makara, G. M.; Keseru, G. M.; Kajtar-Peredy, M.; Anderson, W. K. J. Med. Chem. 1996, 39, 1236-42. (30) Longsworth, L. G. J. Phys. Chem. 1960, 64, 1914-1917. (31) Dawson, R. M. C.; Elliot, D. C.; Elliot, W. H.; Jones, K. M. Data for Biochemical Research, 3rd ed.; Clarendon Press: Oxford, 1986. (32) Sadler, P. J.; Tucker, A.; Viles, J. H. Eur. J. Biochem. 1994, 220, 193200. (33) Janatova, J.; Fuller, J. K.; Hunter, M. J. J. Biol. Chem. 1968, 243, 361222. (34) Buzzeo, M. C., et al. Inorg. Chem. 2004, 43, 7709-25. 9380 J. AM. CHEM. SOC.

9

VOL. 128, NO. 29, 2006

(2)

(assuming a spherical shape for the molecules), where k is the Boltzmann constant (1.38 × 10-16 g cm2 s-2 K-1), T is the temperature of the experiments (303 K), and η is the viscosity of the solvent (MeOH) at 303 K (5.1 × 10-3 g cm-1 s-1). 2.5. Mass Spectrometry. All mass spectra were collected on a Perkin-Elmer API III triple quadrupole instrument (PE-Sciex, Thornhill, Toronto, Canada) fitted with an electrospray interface and operated in the positive ionization mode. Spectra were collected at a low (orifice potential 40 V) or high (orifice potential 100 V) energy. Samples were applied via a direct infusion method at a flow rate of 20uL/min. The temperature of the interface was constant at 40 °C, and nitrogen was used as the nebulizer gas. A minimum of 10 scans was collected for each sample. Spectra were processed using MacSpec software (PE Sciex) on a Macintosh computer. 2.6. Competition with Glycine. The competition between MPA and glycine for Cu(II) was monitored by UV/vis spectroscopy. Additions of glycine were made from an aqueous stock solution of 0.05 M glycine. Glycine has an absolute affinity (pH independent) for Cu(II) of 4.3 × 109 M-1 and a stoichiometry of 2:1 (gly/Cu(II)).31 At pH 6.0 the dissociation constant is ∼16 µM (log R ) pKa - pH ) 9.63-6.0 ) 3.63, log K1(app) (Cu(II)) ) 8.1-3.63 ) 4.47 or 2.95 × 104 M-1).31 2.7. Transfer of Cu(II) to Serum Albumin. The competition of bovine serum albumin (BSA) and MPA for Cu(II) was monitored by EPR. MPA (0.8mM) with 1 equiv of Cu(II) was prepared in 99% H2O/ 1%MeOH with additions of BSA, from a 1.4 mM solution of BSA in H2O, to give a final volume of 1 mL. The pH of all MPA solutions was ∼6.3. The concentration of BSA was based on an A279 (1 mg/mL, 1 cm) ) 0.667.32,33

3. Results and Discussion

3.1. pH Dependence and Cu(II) Binding by UV/visible Spectroscopy. Using ultraviolet and visible absorption spectroscopy we have investigated the Cu(II) binding to MPA and its pH dependence. The pH dependence of MPA in 99% H2O/ 1% MeOH is shown in Figure 1A. As the pH is increased from ∼pH 3 to ∼pH 9 the spectra show a number of changes. The intensity of the three main low pH transitions observed at approximately 225, 250, and 305 nm all decrease as the pH is increased, and new transitions are observed at approximately 235 and 342 nm. The clear isosbestic points provide evidence for a simple equilibrium, which may simply be the presence of only two absorbing species. The intensities of the 305 and 342 nm peaks are plotted as a function of pH and are shown in the inset to Figure 1A. The intensities of these two peaks are equivalent at ∼pH 8.2, which most likely corresponds to the pKa of the phenolic group of MPA. The species in solution are

Copper(II) Binding to Mycophenolic Acid

Figure 1. UV/visible spectra of Cu(II) binding to MPA. (Panel A) pH titration of apo-MPA in 99%H2O/1%MeOH. MPA (0.1 mM) was titrated with K2HPO4 from pH 4.8 to pH 9.2. Inset a shows the absorbance at 305 and 342 nm plotted as a function of pH. (Panel B) Cu(II) titration of 0.1 mM MPA in 100%MeOH. Cu(II) was added in 0.25 mol equiv until a maximum of 2 equiv had been added. Inset b shows the absorbance at 350 nm plotted as a function of mole equivalents of copper added. (Panel C) pH titration of 0.5mM [CuMPA] in 96% H2O/4% MeOH. Inset c shows Cu(II) d-d transitions as a function of pH.

likely to be the carboxylate anion at low pH values and the carboxylate and the phenolate anion at pH values > 8.5. The titration of MPA with Cu(II) in MeOH is shown in Figure 1B. As Cu(II) is titrated into a solution of MPA a new band becomes apparent at ∼350 nm. Weak transitions around 350 nm have been observed in many copper complexes and have been attributed to square-pyramidal [CuO4O] complexes,19 and CuO4 centers in [Cu(OAr)4] complexes.34 In the latter case, the transition is considered d-d in origin. Increases are also observed in the range 230-300 nm, which may be due to Cu(II)-carboxy charge transfer absorption that usually occurs between 240 and 280 nm.35 The inset in Figure 1B shows the absorption at 350 nm plotted as a function of the Cu(II) equivalents. The absorption at 350 nm tends to plateau after 1 equiv of Cu(II) has been added. However, the stoichiometry is difficult to ascertain due to the lack of a distinct point at which the plateau begins, suggesting weak binding or the presence of multiple species. Indeed, the Cu(II) titration (Figure 1B) shows no isosbestic points that would indicate formation of a single copper-bound species. Peaks due to Cu(II) transitions were also observed in the visible region at ∼450 nm (giving the solution a pale yellow (35) Dendrinou-Samara, C.; Jannakoudakis, P. D.; Kessissoglou, D. P.; Manoussakis, G. E.; Mentzafos, D.; Terzis, A. J. Chem. Soc., Dalton Trans. 1992, 22, 3259-3264.

ARTICLES

Figure 2. Mass spectrometric analysis of Cu(II) binding to MPA. (Panel A) Spectrum obtained for MPA (1 mM) at pH 5 in 99%H2O/1%EtOH. The orifice potential was 40 V. Major peaks labeled correspond to MPA ammonium adducts. (Panel B) Spectrum obtained for MPA (1 mM) loaded with 1 mol equiv of Cu(II) at pH 5 (99%H2O/1%EtOH) at low orifice potential (40 V). (Panel C) Same sample as that for Panel B, but spectrum was recorded with a high orifice potential (100 V).

color) and at ∼750 nm. A pH titration of MPA with 1 mol equiv of Cu(II) bound (Figure 1C) shows that the pH affects many of the transitions. As the pH increases, the peaks at 350, 450, and 750 nm all increase in intensity. At pH 8.4 the 350 nm peak overlaps with transitions due to the phenolate anion as observed in Figure 1A. The peak initially at 750 nm (shown inset Figure 1c) shows a steady hypsochromic shift, moving to ∼690 nm at pH 8.4. The λmax of the d-d bands is largely dependent on the nature of the equatorial ligands.36,37 At low pH (pH 5.1) the d-d transitions observed in this work occur at a wavelength near 750 nm. Using the method of Prenesti et al.,38 based on that of Billo37 (eq 1) the predicted λmax of a single Cu(II) bound to two carboxylate groups and two water ligands would be ∼770 nm, which corresponds favorably with the pH 5.1 λmax (∼750 nm) observed in Figure 1c. As the pH increases to around 7, the λmax moves to ∼700 nm which could represent Cu(II) coordinated to carboxylates and a mixture of water and hydroxide ligands. Beyond this pH the λmax moves below 700 nm, suggesting the ligands coordinated to the Cu(II) ion involve carboxylate and either alcoholate or hydroxide moieties. Possibly, the coordination at high pH involves contributions from the phenolate anion. 3.2. Copper-MPA Complexes by Mass Spectrometry. The electrospray mass spectra of apo- and Cu(II)-bound MPA are shown in Figure 2. The spectrum of apo-MPA (Figure 2A) (36) Bryce, G. F.; Gurd, F. R. J. Biol. Chem. 1966, 241, 1439-1448. (37) Billo, R. J. Inorg. Nucl. Chem. Lett. 1974, 10, 613. (38) Prenesti, E.; Daniele, P. G.; Prencipe, M.; Ostacoli, G. Polyhedron 1999, 18, 3233-3241. J. AM. CHEM. SOC.

9

VOL. 128, NO. 29, 2006 9381

Jones et al.

ARTICLES

Figure 3. A 500 MHz NMR analysis of Cu(II) binding to MPA. A small aliquot (∼0.2 mol equiv) of Cu(II) was added to MPA (∼2 mM) in d4-MeOH at 300 K. (Panel A) 1H NMR spectra of apo-MPA (bottom trace) and [CuMPA] (top trace); (Panel B) 13C NMR spectra of apo-MPA (bottom trace) and [CuMPA] (top trace); and (Panel C) schematic of MPA with positions in bold highlighting of 13C atoms perturbed by addition of paramagnetic Cu(II). Resonance assignments are based on those of Makara et al.29 * indicates solvent peaks.

shows that MPA (MW ) 320.3 g/mol) exists in this solution (∼90%H2O/10%EtOH) as a mixture of monomer ([MPA + NH4]+, m/z 338.2), dimer ([MPA2 + NH4]+, m/z 658.4) and some trimer ([MPA3 + NH4]+, m/z 978.3). These masses reflect the formation of ammonium ion adducts as a result of residual contaminating buffer in the electrospray interface. Addition of Cu(II) (as CuCl2) results in the disappearance of the m/z 338.2, 658.4, and 978.3 peaks and the appearance of a new set at predominantly m/z 702.2 and 1022.4, as shown in Figure 2B. The m/z 702.2 peak is most likely two MPA molecules after the loss of two labile carboxylic acid protons and the addition of the Cu(II) ion ([Cu(MPA)2 - 2H]+). The peak at m/z 1022.5 involves formation of [Cu(MPA)2 - 2H]+ followed by addition of a neutral MPA molecule ([Cu(MPA)2 - 2H + MPA]+), although it is unknown if the additional MPA molecule is a copper ligand. A mass spectrum acquired with 2 mol equiv of Cu(II) shows no change to the ratio of the m/z 702.2 and 1022.4 peaks which suggests that the third MPA molecule is not involved as a copper ligand (data not shown). Indeed, association between MPA molecules is observed in the spectrum of apoMPA. Analysis of the d-d transitions in Figure 1c suggested that, at the pH of the mass spectrometry experiments (pH 5), the predominant Cu(II) complex involved carboxylate and water ligands. Analysis of Figure 2B shows a peak at m/z 738.2 representing [Cu(MPA)2 - 2H + 2H2O]+, confirming the UV/ vis analysis. The observation of this peak, even at low orifice potentials (40 V), is unexpected as fragmentation during electrospray ionization can lead to a loss of solvent molecules and, at higher orifice potentials, may even result in ligand dissociation.39,40 The fact that any m/z 738.2 molecular ion is observed is testament to the stability of the Cu(II) complex. 9382 J. AM. CHEM. SOC.

9

VOL. 128, NO. 29, 2006

Further coordination from solvent molecules is also observed at m/z 428.4 ([CuMPA - 2H + C2H5OH]+). Chloride ion does not appear to be involved in any of the major complexes formed. At a very high orifice potential (100 V) the spectrum shows complete loss of all solvent bound complexes and [Cu(MPA)3]+ (Figure 2C), but retention of [Cu(MPA)2]+ and production of a copper bound monomer ([CuMPA - 2H]+, m/z 382.0) and a low abundance binuclear, dimeric species ([Cu2MPA2 - 2H]+, m/z 765.0). This result shows that [Cu(MPA)2]+ is one of the most stable species. The mass spectroscopic data show that there is heterogeneity in the type of Cu(II) complexes formed, providing a rationale as to why the stoichiometry is difficult to determine from the UV/visible data alone. 3.3. Structural Analysis by 1H and 13C NMR. Figure 3A shows the 1H NMR spectra obtained for apo-MPA and after a small amount of Cu(II) has been added. Cu(II) is a d9 ion that is paramagnetic irrespective of the coordination geometry (for mononuclear complexes). The paramagnetic nature of Cu(II) tends to broaden all MPA resonances to some extent; however, some peaks show a greater affect upon Cu(II) binding than others. The broadening of resonances can occur as a result of through-bond (contact) or through-space (pseudo-contact) interactions. Typically, pseudo-contact effects only occur over short distances ( K2. Although this work shows that Cu(II) will be lost to albumin, it is also possible

9386 J. AM. CHEM. SOC.

9

VOL. 128, NO. 29, 2006

that the metal may be lost earlier to enzymes and proteins found in the GI tract. In this respect copper-bound MPA may act similarly to EC-MPS; however, instead of a slowly dissolving coat, the loss of Cu(II) leaves free MPA to exert a clinical effect. This also leaves Cu(II) to either be absorbed and utilized or excreted. The low pH of the stomach would result in greater loss of copper compared to the increased pH of the lower intestine. Indeed, copper-indomethacin has a greater effect in the lower GI tract compared to the stomach, although the copperbound drug was less damaging in both areas than indomethacin alone.20 Gastrointestinal intolerability is one of the limiting factors in optimal MPA use. GI damage can occur due to a number of factors including direct topical irritation to the GI epithelium, impaired barrier properties (e.g., binding to phospholipids), suppression of prostaglandin synthesis, and interference with the repair mechanism after superficial injury.15 In the case of MPA, further GI toxicity may arise due to the antiproliferative effect of the drug on enterocytes.1 Importantly, however, the reduced severity of GI effects observed with ECMPS suggests that a direct topical interaction is a contributing factor. Although much further physiologic investigation is required, the characterization of the copper mycophenolate complex provided here suggests that it may find a place in the arsenal of immunosuppressive therapy. To this end, investigations into the interactions of MPA and copper-bound MPA with synthetic phosphocholine bilayers are currently underway. Acknowledgment. C.E.J. is the recipient of a C.J. Martin Fellowship from the National Health and Medical Research Council of Australia. Dr. Charles Dameron is thanked for the initial supply of MPA, and Dr. Tri Le is thanked for NMR assistance. Supporting Information Available: Figure S1 shows a representative intensity versus NMR gradient field strength plot used to obtain diffusion coefficients of copper-bound MPA. Figure S2 shows the S-band EPR spectrum for copper-bound MPA along with the computer simulation of the experimental spectrum. Figure S3 shows the UV/vis and mass spectrometry competitive metal ion capture experiments conducted with acetic acid. S4 is a table of line width parameters for the major and minor copper-bound MPA simulations shown in Figure 4. S5 lists the full authorship of cited articles with >10 authors. This material is available free of charge via the Internet at http://pubs.acs.org. JA057651L